[VSync Queue] Flush input in CommitComplete()

[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|
  * Also make sure it's outside an impl frame

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGediBEAAJ

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2841263002
Cr-Commit-Position: refs/heads/master@{#469118}
Committed: https://chromium.googlesource.com/chromium/src/+/a237ad2e80ec9ff5cf211b8f28bc2525349e55e0

[chongz, 2017-04-26 22:20:35]

Description was changed from

==========
Only queue input events inside impl frame

BUG=
==========

to

==========
Only queue input events inside impl frame

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-26 22:21:28]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-26 22:22:00]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-27 02:19:28]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-27 02:19:29]

Dry run: Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[chongz, 2017-04-27 16:04:14]

Description was changed from

==========
Only queue input events inside impl frame

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-27 16:14:20]

Description was changed from

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                                               TOT              
Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-27 16:14:35]

Description was changed from

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                                               TOT              
Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT               Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-27 16:14:47]

Description was changed from

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT               Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT             Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-27 16:17:19]

Description was changed from

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT             Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT             Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-04-27 16:19:46]

chongz@chromium.org changed reviewers:
+ sunnyps@chromium.org

[chongz, 2017-04-27 16:19:47]

sunnyps@ PTAL, thanks!

https://codereview.chromium.org/2841263002/diff/1/cc/trees/layer_tree_host_im...
File cc/trees/layer_tree_host_impl.cc (right):

https://codereview.chromium.org/2841263002/diff/1/cc/trees/layer_tree_host_im...
cc/trees/layer_tree_host_impl.cc:1918: input_handler_state_ =
InputHandlerState::IDLE;
Is this the right place, or should I put it in the start of
|LayerTreeHostImpl::DrawLayers()|?

[dtapuska, 2017-04-27 16:24:43]

dtapuska@chromium.org changed reviewers:
+ dtapuska@chromium.org

[dtapuska, 2017-04-27 16:24:44]

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h
File cc/input/input_handler.h (right):

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h#ne...
cc/input/input_handler.h:113: enum class InputHandlerState {
This name seems a bit too generic.

https://codereview.chromium.org/2841263002/diff/1/testing/variations/fieldtri...
File testing/variations/fieldtrial_testing_config.json (right):

https://codereview.chromium.org/2841263002/diff/1/testing/variations/fieldtri...
testing/variations/fieldtrial_testing_config.json:3205: "VsyncAlignedInput": [
The variation change should be a separate change entirely.

[chongz, 2017-04-27 19:09:13]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-27 19:09:41]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[chongz, 2017-04-27 19:12:14]

dtapuska@ PTAL again, thanks!

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h
File cc/input/input_handler.h (right):

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h#ne...
cc/input/input_handler.h:113: enum class InputHandlerState {
On 2017/04/27 16:24:44, dtapuska wrote:
> This name seems a bit too generic.

Changed to READY_FOR_INPUT and BUSY_INSIDE_IMPL_FRAME, or do you have other
suggestions?

https://codereview.chromium.org/2841263002/diff/1/testing/variations/fieldtri...
File testing/variations/fieldtrial_testing_config.json (right):

https://codereview.chromium.org/2841263002/diff/1/testing/variations/fieldtri...
testing/variations/fieldtrial_testing_config.json:3205: "VsyncAlignedInput": [
On 2017/04/27 16:24:44, dtapuska wrote:
> The variation change should be a separate change entirely.

Removed. Was uploaded for telemetry testing.

[commit-bot: I haz the power, 2017-04-27 19:34:41]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-27 19:34:42]

Dry run: Try jobs failed on following builders:
  linux_trusty_blink_rel on master.tryserver.blink (JOB_FAILED,
http://build.chromium.org/p/tryserver.blink/builders/linux_trusty_blink_rel/b...)
  chromeos_daisy_chromium_compile_only_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_daisy_...)
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[chongz, 2017-04-27 22:15:17]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-27 22:30:29]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-28 01:37:41]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-28 01:37:42]

Dry run: Try jobs failed on following builders:
  linux_trusty_blink_rel on master.tryserver.blink (JOB_FAILED,
http://build.chromium.org/p/tryserver.blink/builders/linux_trusty_blink_rel/b...)

[chongz, 2017-04-28 15:00:22]

chongz@chromium.org changed reviewers:
+ danakj@chromium.org

[chongz, 2017-04-28 15:00:23]

danakj@ PTAL, thanks!

[chongz, 2017-04-28 17:54:45]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[chongz, 2017-04-28 17:55:18]

Patchset #2 (id:20001) has been deleted

[commit-bot: I haz the power, 2017-04-28 17:55:26]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[chongz, 2017-04-28 19:21:21]

dtapuska@ I've renamed enum |InputEventState| to |ImplThreadPhase| according to
our offline discussion. PTAL again, thanks!

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h
File cc/input/input_handler.h (right):

https://codereview.chromium.org/2841263002/diff/1/cc/input/input_handler.h#ne...
cc/input/input_handler.h:113: enum class InputHandlerState {
On 2017/04/27 19:12:14, chongz wrote:
> On 2017/04/27 16:24:44, dtapuska wrote:
> > This name seems a bit too generic.
> 
> Changed to READY_FOR_INPUT and BUSY_INSIDE_IMPL_FRAME, or do you have other
> suggestions?

Changed enum name to |ImplThreadPhase|.

[commit-bot: I haz the power, 2017-04-28 19:34:22]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-28 19:34:23]

Dry run: This issue passed the CQ dry run.

[chongz, 2017-05-01 19:14:23]

dtapuska@ Can you take a look at the new enum name please? Thanks!

[dtapuska, 2017-05-01 19:24:32]

On 2017/05/01 19:14:23, chongz wrote:
> dtapuska@ Can you take a look at the new enum name please? Thanks!

lgtm. That name is definitely better than InputEventState.

[chongz, 2017-05-01 19:26:33]

sunnyps@ danakj@ PTAL at cc implementation, thanks!

[chongz, 2017-05-03 15:55:25]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-05-03 15:55:50]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[chongz, 2017-05-03 16:10:36]

Description was changed from

==========
[VSync Queue] Only queue gesture events during impl frame

Queuing events all the time will cause checkerboard issue in high latency mode.

This CL changed the process to:
1. Queue gesture events between |LayerTreeHostImpl::WillBeginImplFrame()| and
    |LayerTreeHostImpl::DidFinishImplFrame()|.
2. Otherwise flush and forward gesture events immediately.
3. Still flush gesture events on |LayerTreeHostImpl::WillBeginImplFrame|.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

Patch 1  mac_retina_perf_bisect test result: （https://goo.gl/bpXHie）
                                    TOT             Patch
mean_pixels_checkerboarded          260.260%        223.180%

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all
input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[chongz, 2017-05-03 16:12:01]

Description was changed from

==========
[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all
input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all
input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|
  * Also make sure it's outside an impl frame

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[commit-bot: I haz the power, 2017-05-03 17:16:29]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-05-03 17:16:30]

Dry run: This issue passed the CQ dry run.

[danakj, 2017-05-03 17:20:13]

LGTM

[sunnyps, 2017-05-03 18:37:14]

lgtm

https://codereview.chromium.org/2841263002/diff/60001/cc/trees/layer_tree_hos...
File cc/trees/layer_tree_host_impl.cc (right):

https://codereview.chromium.org/2841263002/diff/60001/cc/trees/layer_tree_hos...
cc/trees/layer_tree_host_impl.cc:344: if (input_handler_client_ &&
impl_thread_phase_ == ImplThreadPhase::IDLE)
nit: can you please add a comment explaining why we're doing this here?

[chongz, 2017-05-03 19:31:01]

The CQ bit was checked by chongz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-05-03 19:31:33]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[chongz, 2017-05-03 20:32:33]

https://codereview.chromium.org/2841263002/diff/60001/cc/trees/layer_tree_hos...
File cc/trees/layer_tree_host_impl.cc (right):

https://codereview.chromium.org/2841263002/diff/60001/cc/trees/layer_tree_hos...
cc/trees/layer_tree_host_impl.cc:344: if (input_handler_client_ &&
impl_thread_phase_ == ImplThreadPhase::IDLE)
On 2017/05/03 18:37:13, sunnyps wrote:
> nit: can you please add a comment explaining why we're doing this here?

Done.

[commit-bot: I haz the power, 2017-05-03 20:56:32]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-05-03 20:56:33]

Dry run: This issue passed the CQ dry run.

[chongz, 2017-05-03 20:59:05]

The CQ bit was checked by chongz@chromium.org

[chongz, 2017-05-03 20:59:06]

The patchset sent to the CQ was uploaded after l-g-t-m from
dtapuska@chromium.org, danakj@chromium.org, sunnyps@chromium.org
Link to the patchset: https://codereview.chromium.org/2841263002/#ps80001
(title: "sunnyps's review: Add comment")

[commit-bot: I haz the power, 2017-05-03 20:59:35]

CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-05-03 21:05:36]

CQ is committing da patch.

Bot data: {"patchset_id": 80001, "attempt_start_ts": 1493845145888080,
"parent_rev": "d4b5dbc19c72bfabba394a7e728fd67b96d7287b", "commit_rev":
"a237ad2e80ec9ff5cf211b8f28bc2525349e55e0"}

[commit-bot: I haz the power, 2017-05-03 21:05:47]

Description was changed from

==========
[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all
input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|
  * Also make sure it's outside an impl frame

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
[VSync Queue] Flush input in CommitComplete()

We should call |DeliverInputForBeginFrame()| in two places:
1. Low latency mode (begin main frame and commit every frame)
  * Flush input in |WillBeginImplFrame()| and the commit should include all
input
2. High latency mode (begin main frame and commit are late)
  * Flush input in |CommitComplete()| and before |PrepareTiles()|
  * Also make sure it's outside an impl frame

Tested locally and checkerboard issue seems to be fixed.

Graphics-dev discussion:
https://groups.google.com/a/chromium.org/d/msg/graphics-dev/xOdp3hIxZvg/SQtGe...

BUG=704732
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2841263002
Cr-Commit-Position: refs/heads/master@{#469118}
Committed:
https://chromium.googlesource.com/chromium/src/+/a237ad2e80ec9ff5cf211b8f28bc...
==========

[commit-bot: I haz the power, 2017-05-03 21:05:49]

Committed patchset #4 (id:80001) as
https://chromium.googlesource.com/chromium/src/+/a237ad2e80ec9ff5cf211b8f28bc...